Mapping wetland vegetation using Sentinel-1 and SWOT data
Bibliographic record
Abstract
Flooded vegetation mapping is critical for disaster response, ecological monitoring, wetland mapping and effective water resources management, yet it remains a significant challenge due to the limitations of optical remote sensing in detecting inundation beneath dense canopies and cloud cover. Monitoring of the dynamics of wetlands and flooded areas can help improve hydrological modeling and forecasting by providing accurate and timely information on water storage within a watershed. This study presents a methodology for the automated segmentation and classification of flooded vegetation by combining Sentinel-1 Synthetic Aperture Radar (SAR) data with Surface Water and Ocean Topography (SWOT) mission observations. Previous research has demonstrated the potential of Sentinel-1 SAR for mapping wetlands and delineating flood extents. The proposed method leverages the complementarity between high-resolution spatial information from Sentinel-1 and hydrodynamic information provided by SWOT to map flooded vegetation with an object-based image analysis (OBIA) approach. The method is tested on two study areas, one in the Oromocto River basin in New Brunswick, Canada (45.78 N, 66.55 W) and the other is located around the Mamawi Lake in northern Alberta, Canada (58.66 N, 111.50 W). The first area focuses on wetlands along the Oromocto River which are prone to flooding during the spring freshet and heavy rainfall events. The second area is located within the Peace-Athabasca Delta, which is a complex system of interconnected lakes and wetlands and one of the current test sites for the SWOT and NORthern laKeS (SNORKS) project. Sentinel-1 Interferometric Wide images are selected in alignment with SWOT overpass dates within a 48-hour interval. Observations of high-water levels (including a major flood event for Oromocto) and low-water levels are used to test the algorithm. Preprocessing is applied to Sentinel-1 Ground Range Detected images comprised of radiometric calibration, terrain correction and speckle filtering on both VV and VH polarisations. An image segmentation based on the Mean Shift algorithm is then performed on the dual band images (VV and VH) using the Orfeo Toolbox to create polygons with uniform backscattering behaviour. The SWOT Raster products are then sampled within the segmented polygons to extract statistics of Water Surface Elevation (WSE) and Water Fraction (WF). All polygons with WF greater than 70% and WSE quality rating of 0 or 1 are considered flooded or open water. The segmentation outputs will be validated against flood extent maps provided by Natural Resources Canada where available, as well as maps produced by visual interpretation. Performance is benchmarked using standard metrics such as overall accuracy, class-specific IoU, precision, and recall. Maps produced by visual interpretation are also validated with in situ water gauges to validate water levels measured at the time the images were taken, along with high-resolution Digital Terrain Models (LiDAR), which are used to estimate water extents. The proposed method provides: • High-resolution, temporally consistent maps of flooded vegetation, supporting hydrological and hydraulic modelling and long-term environmental monitoring. • Enhanced understanding of flood propagation dynamics in vegetated landscapes and wetlands to help improve hydrological assessments. • A transferable methodological framework that leverages the complementary strengths of Sentinel-1 SAR and SWOT data, facilitating scalable flood mapping in diverse geographic and climatic contexts. By proposing a simple, automated, and robust method for the detection and classification of inundated vegetation, this project advances the state of flood mapping science and provides critical information for applications in hydrology, hydrodynamics and environmental monitoring, especially in remote areas.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".